iRay Convergence

gooHero

Hey!

I have found my answer online on what "iRay convergence" means. I am still confused as to why some renders start out with, for example, 48% though.

How is that possible?

One more thing... Renders in a fairly dark environment makes it harder for iRay to calculate where the light needs to go and which pixels to light up and what to bounce off of and so on, right? but when I add more light to a scene, in some cases, it makes no difference in render time at all. How come?

So it makes no difference if you up the lumen, but it would make a difference adding another light source?

It's hard to maintain a fair amount of realism in scenes, when you have those nonsensical light sources all over the place (fx facelights and point lights, with no physical light sources near them).

Do you guys have any tricks in that regard? (I already use ghost lights and they are the ones I am referring to)

Kind regards and Happy Holidays,

anti

Richard Haseltine

Iray runs a few iterations before the first convergence report, it is possible that some simple areas will be converged by then.

It isn't the amount of light that matters, much (a brighter source may bounce a few more tiems before stopping, which may help a little) but the complexity of the paths to reach all areas of the image - the more convoluted the path that needs to be followed (mutliple bounces) the longer it will take for a pixel to accumulate enough hits to achieve a settled value. Adding more lights, in different locations (or with a gerater source area) the sooner the image will achieve an acceptable degree of convergence.

One thing to try is helping the process along by adding a light source at a point from which light will eventually bounce, rather than jusdt a free-floating light source with no relation to the physics of the scene.

gooHero

Richard Haseltine said:

Iray runs a few iterations before the first convergence report, it is possible that some simple areas will be converged by then.

It isn't the amount of light that matters, much (a brighter source may bounce a few more tiems before stopping, which may help a little) but the complexity of the paths to reach all areas of the image - the more convoluted the path that needs to be followed (mutliple bounces) the longer it will take for a pixel to accumulate enough hits to achieve a settled value. Adding more lights, in different locations (or with a gerater source area) the sooner the image will achieve an acceptable degree of convergence.

One thing to try is helping the process along by adding a light source at a point from which light will eventually bounce, rather than jusdt a free-floating light source with no relation to the physics of the scene.

Thank you! I tried to make a scene blanketed in light sources, and it kind of works. A full HD render animation test cycle was done in the same time a normal HD render cycle were usually done. The difference in time consumption from scene to scene, with the same characters, but similar primitive environments and light sources, still baffles me though. I think, at this point, I need a new graphics card for 3D stuff to make a real dent in render times. 8Gb VRAM just sucks. I saw that the 4090 was 149% faster than my 3070 at everything - NEED! xD

regards,

anti